A Comprehensive Approach To Assess Current and Future Vulnerabilities In The Columbia River Reservoir System Under Climate Change.

The water resources of the Columbia River Reservoir System sustain an economically critical food-energy-water nexus. Across the system, long-term water management planning is challenged by future hydrologic variability associated with climate change. The US Army Corps of Engineers has established climate policy aimed to reduce operational vulnerability to the effects of climate change and variability; however, the use of climate information in policy and planning continues to be developed and is typically limited to qualitative assessments due to large uncertainties associated with projected outcomes. Here, we present a novel approach for quantitatively demonstrating the performance of current operating criteria of the Columbia River Reservoir System under a wide range of historical and future hydrological conditions and further attributing shifts in performance to underlying hydrologic and operational conditions. A 160-member ensemble of hydrologic projections derived from Coupled Model Intercomparison Project version 5 Global Climate Models is used to derive ensembles of hydro-regulated flow conditions and a suite of hydrologic and operational metrics that could explain performance degradation. Shifts in the frequency of operational vulnerabilities, defined as the inability to meet an operational objective or flow condition, are identified. A machine learning algorithm Random Forest is applied to attribute operational vulnerabilities to the most influential hydrologic and operational conditions. Finally, the relative frequency of reaching these conditions is projected through time. This attribution approach provides a comprehensive characterization of the vulnerability to support hydrological monitoring and design of adaptive resource management strategies.